GEOGRAPHIC CODING Re: The WLS questionnaire collected the zip code of respondent's current address but did not obtain the county of residence. Various data are available from the government and elsewhere at the county level, or for aggregations of counties such as Metropolitan Statistical Areas (MSA's) and Labor Market Areas (LMA's). In order to make it possible to use such data in conjunction with WLS data, the zip codes were mapped to the counties in which they predominantly lie. The non-public-release version of the data contains the resulting county codes, which are called 5-digit FIPS codes, as well as corresponding MSA and LMA codes. This document explains how these codes were created. The relevant variables are as follows. (Marginal frequencies are found in the main codebook listing.) None are available on the public release tape. 58-62 RA021RE Five digit FIPS code for county containing primary respondent's 1992-1993 zip code. 63-67 RA022RE 1990 based LMA code for zip code of primary respondent's 1992-1993 address. The first two digits are the Labor Market Area codes and the last two are commuting zones within the LMA. 68-72 RA023RE 1980 based LMA code for zip code of primary respondent's 1992-1993 address. 73-76 RA024RE 1993 Metropolitan Statistical Area code for primary respondent's 1992-1993 address. 77-83 RA025RE 1990 county population based on primary respondent's zip code of 1992-1993 address. I. ZIP CODE TO COUNTY MAPPING, AND ITS ACCURACY Zip codes cross county boundaries, sometimes, making it necessary in practice to map the zip codes to the counties in which they predominantly lie. If most zip codes lie entirely or almost entirely within single counties, this overlapping of codes and counties should make little difference for social scientific analysis. A file of zip code to primary county code mappings was obtained from the CIESIN archive, at: ftp ftp.ciesin.org login: anonymous cd /pub/census cd usa/stf/0code The name of the file held in this location is zipcnty.sas, which may be obtained with the ftp command "get zipcnty.sas". The file is a SAS format statement identifying for each zip code the 5-digit FIPS code of the county which forms the largest intersection with the zip code's area. FIPS (Federal Information Processing Standards) codes are listed for all counties in FIPS Publication 55 (available at Wendt Library on campus). A FORTRAN program was used to strip the information from the SAS format statement and create a fixed-field file of code correspondences. The county names and associated FIPS codes for those counties in Wisconsin are in cor732a.asc as of 04/05/01. The creator of the source SAS format file, I believe, was: John Blodgett Urban Information Center / Office of Computing University of Missouri - St. Louis 8001 Natural Bridge Rd. St. Louis, Mo 63121-4499 Phone: (314) 516-6014/6000 FAX: 516-6007 Blodgett describes the origins of the SAS file in file "zipread.me" located in the same "Ocode" subdirectory referenced above. His source was a U.S. Postal Service file, the quarterly "city-state" file, and he believes the version used was from April, 1994. Blodget also did an analysis of the county-zip code overlap problem, using the 1990 Census Bureau "ZIP-Block Equivalancy files" (aka "STF3B Headers"). He calculated for each of the 29,472 zip codes the proportion of the zip code area that was within the county which contained the largest amount of its area, and also the proportion of the zip code which was in a metropolitan statistical area. He presents this table cross-classifying these two features of zip codes: Pct of ZIP Code in Primary County by Metro/Non-metro Status Source: 1990 STF3B ZIP Equivalency Files (Filetype=zipeq90x) TABLE OF PCTCO1 BY MNM PCTCO1(% ZIP in primary co.) MNM(Metro/non-metro) Frequency| Percent |In mixed|In metro|Nonmetro| | area | area | area | Total ---------+--------+--------+--------+ 0 to 50 | 29 | 23 | 65 | 117 | 0.10 | 0.08 | 0.22 | 0.40 ---------+--------+--------+--------+ 50 to 67 | 169 | 168 | 251 | 588 | 0.57 | 0.57 | 0.85 | 2.00 ---------+--------+--------+--------+ 67-90 | 371 | 316 | 617 | 1304 | 1.26 | 1.07 | 2.09 | 4.42 ---------+--------+--------+--------+ 90-99.9+ | 138 | 643 | 882 | 1663 | 0.47 | 2.18 | 2.99 | 5.64 ---------+--------+--------+--------+ 100% | 32 | 11277 | 14491 | 25800 | 0.11 | 38.26 | 49.17 | 87.54 ---------+--------+--------+--------+ Total 739 12427 16306 29472 2.51 42.17 55.33 100.00 About 12.5% of zip codes cross county lines, and only 2.5% cross MSA lines. 93% of zip codes have at least 90% of their area in a single county. For most social analyses this seems good enough to attribute the primary county's characteristics to the household on the basis of the zip code. II. LMA CODES. Labor Market Area (LMA) codes were defined as aggregates of counties under a Department of Agriculture project, documented in Tolbert, Charles, and Molly Killian. 1987. "Labor Market Areas for the United States." Washington, DC: Department of Agriculture, Economic Research Service, Agricultural and Rural Economics Division. Tolbert posted an equivalency file for 5-digit county FIPS and Labor Market Area codes at: Gopher: GOPHER.LAPOP.LSU.EDU World Wide Web: http://www.lapop.lsu.edu I obtained the file, czlma903.eqv, and merged some of its fields with a reformatted (fixed columns) version of CIESIN's zipcnty.sas file, the zip-county equivalence file discussed above. The result, zip2lma.dat, has 23541 records mapping zip code ranges to county FIPS codes and the LMA codes, and has been posted in the same "0code" subdirectory on the CIESIN archive referenced above. (There are fewer zip codes than in Blodget's table because of the exclusion of strictly business zip codes and P.O. box codes.) There are two versions of the LMA codes, based on 1980 and 1990 Census commuting data, respectively. Tolbert reported that the documentation on the 1990-standard codes was still incomplete in July of 1995. The WLS data file contains both codes for each respondent, as variables RA022RE (1990) and RA023RE (1980). Variables RA024RE, 1993 Metropolitan Statistical Area code for primary respondent's 1992-1993 address, and RA025RE, 1990 county population based on primary respondent's zip code of 1992-1993 address, were also obtained from Tolbert's county-LMA file. The accuracy of both variables depends on the zip code to county mapping, discussed above.